How to run OCR on a PDF or image with OnyxPDF
- Drop a scanned PDF or an image (photo of a page works too) into the tool.
- Choose the document’s language so the right recognition pack is used.
- Click “Run OCR” and wait while your device reads the pages.
- Copy the text or download it as a .txt file.
Why do it in your browser instead of uploading?
- No upload: OCR of an ID card, letter or contract happens on your own device — CamScanner syncs scans to its cloud and charges for OCR export.
- Seven languages built in, including Urdu and Arabic, which most free web OCR tools do not offer at all.
- Works on both scanned PDFs and plain images of documents.
- Free and unlimited — no per-page OCR caps like Sejda’s free 10-page OCR limit.
Good to know (honest limits)
- Recognition quality depends on the source: sharp, straight, well-lit scans read well; blurry or skewed photos read worse. Scan Enhance can clean a photo up first.
- The vendored “fast” language packs favour speed over the last few percent of accuracy.
- Handwriting is not reliably recognised — this is print OCR.
- The engine and language pack download once (a few MB) and are then cached for offline use; first run on a slow phone takes a little longer.
Frequently asked questions
Which languages does OCR support?
English, Urdu, Arabic, Hindi, Spanish, French and German. Pick the language of your document before running OCR — it noticeably improves accuracy.
Can OCR read a photo of a document?
Yes. A straight, well-lit photo works; for a skewed photo, run Scan Enhance first to straighten and clean it, then OCR the result.
Is my document uploaded for OCR?
No. The Tesseract engine runs as WebAssembly in your browser. Nothing is sent anywhere — unusual for OCR, and the whole point of this site.
Why is the first OCR run slower?
The engine and your chosen language pack load once (a few MB) and are cached on your device. After that, runs start immediately and even work offline.